Physical CAN layer question about Wire Awg

I have a question about signal integrity regarding a CAN cable routing.

I am routing a fairly long run with a few connectors and it is convenient to use 16awg for CAN Hi and Can Lo about half-way through the routing.

Component A ----20awg CAN Hi/Can Lo ---Connector 1
Connector 2 ---16awg Can Hi/Can Lo ---ComponentB

What kind of impact will this have on the signal strength? Let's assume the cable length isn't in question here.

In other words, what does increasing the wire awg mid way through a CAN bus do to the overall signal integrity?

Thanks in advance.

Transmission line effects/reflections

dknguyen said:

Transmission line effects/reflections

Click to expand...

I understand the transmission line is a critical part of the signal integrity, but I am curious on the exact effect when you change the AWG mid-way through a line via a connector. Has this ever been studied? Any articles? ISO Specs? Recommendations? Just curious, I am fairly new to CAN layer in general.

Maybe you didn't understand what I was getting at, but changing the AWG introduces an impedance discontinuity. This discontinuity is a point of reflection. This also holds true for connectors along the line even when the AWG on both sides of the connector is the same. So the answer is that it is the same as any other transmission line impedance discontinuity.

dknguyen said:

Maybe you didn't understand what I was getting at, but changing the AWG introduces an impedance discontinuity. This discontinuity is a point of reflection. This also holds true for connectors along the line even when the AWG on both sides of the connector is the same. So the answer is that it is the same as any other transmission line impedance discontinuity.

Click to expand...

Thanks for this information, it has been helpful. Let me clarify a little further to continue this discussion. I am not disagreeing with you, just want to look further into this discussion.

Ok, changing the AWG introduces an impedance discontinuity, but does this impact it to the point where it is not recommended?

Is there guidelines standards that we can look at?

For instance, we have to use a connector. Many CAN layers consist of connectors, you can't get around that. This is where termination resistors come to play.

My main question / concern is, when is it too much?

Is changing the AWG from 20 to 16 a big no no? Not recommended?

On a side note, does anyone have any good books or classes they can recommend to go over CAN hardware / software design?

Too much would be when the transmission stops working reliably or if the signal on the scope looks dirtier than you would like. CAN is half-duplex with multiple reception and transmission points so I can't think of a way where you can practically do something about discontinuities other than slowing down your signal edges which will ultimately limit your data rates (which will require transceivers that allow you to do so).

I read somewhere that you can get around discontinuities by providing both source and load termination and drive signal at double what is required at the receiver. Apparently it allows the signal to be received cleanly and supresses from the reflection from the discontinuity from bouncing back and forth and eventually appearing at the receiver. But that requires a lot of control over the system and terminations everywhere on a multi-master bus (that's assuming it even works on a bi-directional line which I don't think it does).

A change in wire size does not automatically cause an impedance discontinuity - different size cables can have the same impedance, or cables with the same copper size can have different impedances.

As long as the two cables have the same impedance (or near enough so, no more than about 1.5:1 ratio), it should not matter.

Examples - 22awg and 28awg twisted pairs suitable for RS485 or CAN etc., both 120 Ohm:
https://uk.rs-online.com/web/p/twisted-multipair-industrial-cable/7491627/
https://uk.rs-online.com/web/p/twisted-multipair-industrial-cable/3319098/

Depends on bitrate & pull up resistors used, but probably not much effect.
I'm sure you already know for class A you need twisted wires for hi & lo.